Multiphase Bioreactor Design

The quantitative investigation of substrate uptake requires a time resolution of several
few 100 ms, otherwise, artifacts must result. One can stop glucose metabolism within 100
ms by spraying the cell suspension from the over-pressurized bioreactor into 60%
methanol which was pre-chilled to −40°C. This procedure does not damage the integrity
of the cells and the pre-treated sample remains liquid which is advantageous for further
processing.

Trends in sensor development

Clearly, the development of reliable on-line estimation of intracellular components is a
(the) major challenge for the future (e.g. Schuster, 1999 and other contributions in
Sonnleitner, 1999).

Data Acquisition

All stages in a data acquisition system are represented schematically in Figure 3.2.
Transducers incorporated in measurement devices normally deliver a continuous
electrical signal. This will generally undergo some form of conditioning (galvanic
insulation, amplification, attenuation, etc.) depending on aspects related to transmission
(continuous or digital transmission, distance of transmission, etc.). The final stage is
normally local signal acquisition, corresponding to multiplexing, analog-to-digital
conversion (ADC) and storage in the process computer or front-end-system. Data are
stored in a computer in distinct time intervals resulting in discretisation with respect to
time and concomitantly in a possible loss of information. A data reduction algorithm
should therefore be applied which must account for this fact: raw data should be scanned
with high frequency and the essential data (corresponding only to important changes)
may be stored with the necessary frequency, i.e. variably or not equidistant with respect
to time.
Automated measurement and control of bioprocesses, presently an art but a routine in
the near future, generates a tremendous amount of data. This requires judging of the
importance of these data for documentation in order to implement an effective data
reduction method without loss of valuable information. The reduction algorithm must
keep a true image of the real data. We adopted a simple algorithm to achieve this goal: all
data—independent of whether measured or calculated—are treated as variables and kept
in a circular buffer in the frequency with which they were generated. Values (data-points)
of variables that change significantly with time are written to the archive. The
significance of a change is judged by a reasonably defined window for each variable—
including all intended culture parameters—the width of which is usually determined by
the noise on the respective signal. At any instance, every “first” data-point (of an
experiment) is archived, together with a time-stamp. A next entry to the archive is made
only if the variable moves outside the respective window which had been centered
around the last archived value;

New methodologies for multiphase bioreactors 3 65